Module:SMWUtils

Documentation for this module may be created at Module:SMWUtils/doc

-- Module:SMWUtils
-- Utility functions for Semantic MediaWiki queries
-- Uses frame:preprocess() since Semantic Scribunto is not installed

local p = {}

local function trim(s)
    if s == nil then return nil end
    s = tostring(s)
    s = s:gsub('^%s+', ''):gsub('%s+$', '')
    if s == '' then return nil end
    return s
end

--- Find all results where a chained property equals its minimum value
-- Uses a two-pass approach: first find min, then query for all matching
-- @param frame Frame object with args:
--   conditions: SMW query conditions (e.g., "[[Has video signal format::Component (YPbPr)]]")
--   minProperty: Property to find minimum of (e.g., "Has CRT model.Has screen size inches")
--   printouts: Comma-separated list of properties to display
--   format: Output format (table, ul, ol) - default "table"
--   headers: Comma-separated header names for table format
--   intro: Optional intro text
-- @return Wikitext output
function p.findAllWithMin(frame)
    local args = frame.args
    local conditions = trim(args.conditions) or ''
    local minProperty = trim(args.minProperty)
    local printouts = trim(args.printouts) or ''
    local format = trim(args.format) or 'table'
    local headers = trim(args.headers) or ''
    local intro = trim(args.intro)
    
    if not minProperty then
        return '<span class="error">Error: minProperty is required</span>'
    end
    
    -- Step 1: Find the minimum value using a sorted query with limit=1
    local minQueryWikitext = '{{#ask:' .. conditions .. '\n' ..
        '|?' .. minProperty .. '\n' ..
        '|sort=' .. minProperty .. '\n' ..
        '|order=asc\n' ..
        '|limit=1\n' ..
        '|format=broadtable\n' ..
        '|headers=hide\n' ..
        '|link=none\n' ..
        '}}'
    
    local minResult = frame:preprocess(minQueryWikitext)
    
    -- Parse the minimum value from the table result
    local minValue = nil
    if minResult then
        -- Extract from last <td> in the row (the property value)
        for cellContent in minResult:gmatch('<td[^>]*>(.-)</td>') do
            local cleanContent = cellContent:gsub('<[^>]+>', '')
            local numVal = tonumber(trim(cleanContent))
            if numVal then
                minValue = numVal
            end
        end
    end
    
    if minValue == nil then
        return "No results found or could not determine minimum value."
    end
    
    -- Step 2: Query for all results matching this minimum value
    local printoutList = {}
    for prop in printouts:gmatch('[^,]+') do
        table.insert(printoutList, '|?' .. trim(prop))
    end
    
    local fullQueryWikitext = '{{#ask:' .. conditions .. '\n' ..
        '[[' .. minProperty .. '::' .. minValue .. ']]\n' ..
        table.concat(printoutList, '\n') .. '\n' ..
        '|sort=' .. minProperty .. '\n' ..
        '|limit=100\n' ..
        '|format=broadtable\n' ..
        '|headers=hide\n' ..
        '|link=all\n' ..
        '}}'
    
    local fullResult = frame:preprocess(fullQueryWikitext)
    
    -- Parse results from HTML table
    local results = p._parseTableResults(fullResult, printouts)
    
    if #results == 0 then
        return "No results found matching minimum value " .. minValue .. "."
    end
    
    -- Build output
    local headerList = {}
    for h in headers:gmatch('[^,]+') do
        table.insert(headerList, trim(h))
    end
    
    local printoutProps = {}
    for prop in printouts:gmatch('[^,]+') do
        table.insert(printoutProps, trim(prop))
    end
    
    local introText = intro or ("Smallest screen size with this input: '''" .. minValue .. "\"'''")
    
    if format == 'table' then
        return p._formatTableFromParsed(results, printoutProps, headerList, introText)
    elseif format == 'ul' or format == 'ol' then
        return p._formatListFromParsed(results, format, introText)
    else
        return p._formatTableFromParsed(results, printoutProps, headerList, introText)
    end
end

--- Find all results where a property equals its maximum value
function p.findAllWithMax(frame)
    local args = frame.args
    local conditions = trim(args.conditions) or ''
    local maxProperty = trim(args.maxProperty)
    local printouts = trim(args.printouts) or ''
    local format = trim(args.format) or 'table'
    local headers = trim(args.headers) or ''
    local intro = trim(args.intro)
    
    if not maxProperty then
        return '<span class="error">Error: maxProperty is required</span>'
    end
    
    -- Step 1: Find the maximum value using a sorted query with limit=1
    local maxQueryWikitext = '{{#ask:' .. conditions .. '\n' ..
        '|?' .. maxProperty .. '\n' ..
        '|sort=' .. maxProperty .. '\n' ..
        '|order=desc\n' ..
        '|limit=1\n' ..
        '|format=broadtable\n' ..
        '|headers=hide\n' ..
        '|link=none\n' ..
        '}}'
    
    local maxResult = frame:preprocess(maxQueryWikitext)
    
    -- Parse the maximum value from the table result
    local maxValue = nil
    if maxResult then
        for cellContent in maxResult:gmatch('<td[^>]*>(.-)</td>') do
            local cleanContent = cellContent:gsub('<[^>]+>', '')
            local numVal = tonumber(trim(cleanContent))
            if numVal then
                maxValue = numVal
            end
        end
    end
    
    if maxValue == nil then
        return "No results found or could not determine maximum value."
    end
    
    -- Step 2: Query for all results matching this maximum value
    local printoutList = {}
    for prop in printouts:gmatch('[^,]+') do
        table.insert(printoutList, '|?' .. trim(prop))
    end
    
    local fullQueryWikitext = '{{#ask:' .. conditions .. '\n' ..
        '[[' .. maxProperty .. '::' .. maxValue .. ']]\n' ..
        table.concat(printoutList, '\n') .. '\n' ..
        '|sort=' .. maxProperty .. '\n' ..
        '|limit=100\n' ..
        '|format=broadtable\n' ..
        '|headers=hide\n' ..
        '|link=all\n' ..
        '}}'
    
    local fullResult = frame:preprocess(fullQueryWikitext)
    
    -- Parse results from HTML table
    local results = p._parseTableResults(fullResult, printouts)
    
    if #results == 0 then
        return "No results found matching maximum value " .. maxValue .. "."
    end
    
    -- Build output
    local headerList = {}
    for h in headers:gmatch('[^,]+') do
        table.insert(headerList, trim(h))
    end
    
    local printoutProps = {}
    for prop in printouts:gmatch('[^,]+') do
        table.insert(printoutProps, trim(prop))
    end
    
    local introText = intro or ("Largest screen size with this input: '''" .. maxValue .. "\"'''")
    
    if format == 'table' then
        return p._formatTableFromParsed(results, printoutProps, headerList, introText)
    elseif format == 'ul' or format == 'ol' then
        return p._formatListFromParsed(results, format, introText)
    else
        return p._formatTableFromParsed(results, printoutProps, headerList, introText)
    end
end

-- Internal: Parse HTML table results into a Lua table
function p._parseTableResults(htmlResult, printouts)
    local results = {}
    
    if not htmlResult then
        return results
    end
    
    -- Count expected columns from printouts
    local numPrintouts = 0
    for _ in printouts:gmatch('[^,]+') do
        numPrintouts = numPrintouts + 1
    end
    
    -- Match each table row
    for rowContent in htmlResult:gmatch('<tr[^>]*>(.-)</tr>') do
        local cells = {}
        -- Extract content from each <td>, preserving HTML for links
        for cellContent in rowContent:gmatch('<td[^>]*>(.-)</td>') do
            table.insert(cells, cellContent)
        end
        
        -- First cell is usually the page link, rest are printout values
        if #cells >= 1 then
            table.insert(results, cells)
        end
    end
    
    return results
end

-- Internal: Format parsed results as a wikitable
function p._formatTableFromParsed(results, printouts, headers, introText)
    local output = {}
    
    table.insert(output, introText .. "\n\n")
    table.insert(output, '{| class="wikitable sortable"\n')
    
    -- Header row
    table.insert(output, '|-\n! #')
    for i, prop in ipairs(printouts) do
        local headerText = headers[i] or prop:gsub('.*%.', ''):gsub('Has ', '')
        table.insert(output, ' !! ' .. headerText)
    end
    table.insert(output, '\n')
    
    -- Data rows
    for i, row in ipairs(results) do
        table.insert(output, '|-\n| ' .. tostring(i))
        for j = 1, #printouts do
            local cellVal = row[j] or ''
            table.insert(output, ' || ' .. cellVal)
        end
        table.insert(output, '\n')
    end
    
    table.insert(output, '|}')
    
    return table.concat(output)
end

-- Internal: Format parsed results as a list
function p._formatListFromParsed(results, listType, introText)
    local output = {}
    local bullet = listType == 'ol' and '#' or '*'
    
    table.insert(output, introText .. "\n\n")
    
    for _, row in ipairs(results) do
        local parts = {}
        for _, cell in ipairs(row) do
            if cell and cell ~= '' then
                table.insert(parts, cell)
            end
        end
        table.insert(output, bullet .. ' ' .. table.concat(parts, ' – ') .. '\n')
    end
    
    return table.concat(output)
end

return p